Inhibition of Aromatase

View metadata, citation and similar papers at core.ac.uk brought to you by CORE

provided by Elsevier - Publisher Connector

FEBS Letters 409 (1997) 396-400 FEBS 18708

A novel activity for a group of sesquiterpene lactones : inhibition of aromatase

Javier G. Blanco3*, Roberto R. Gilb, Cecila I. Alvareza, Luis C. Patritoa, Susana Genti-Raimondia, Alfredo Flurya 0Departamento de Bioquimica Clinica - IMBIV (CONICET), Facultad de Ciencias Quimicas, Universidad Nacional de Cordoba, Agenda Postal 4, CC 61, 5000 Cordoba, Argentina hDepartamento de Quimica Orgânica - IMBIV (CONICET), Facultad de Ciencias Quimicas, Universidad Nacional de Cördoba, 5000 Cordoba, Argentina

Received 28 April 1997

agents. However, few metabolites with significant aromatase Abstract A group of eleven sesquiterpene lactones isolated from different Asteraceae species from north-western Argentina were inhibitory activity have been previously described [3]. The investigated for their inhibitory action on the estrogen biosynth- sesquiterpene lactones are a structurally diverse and extremely esis. Seven of them, of different skeleton types, were found to large group of metabolites with over 3500 structures thus far inhibit the aromatase enzyme activity in human placental reported. Over 90% of them have been found in the Astera- microsomes, showing IC50 values ranging from 7 to 110 uM. ceae, and also occur in the Acanthaceae, Anacardiaceae, The most active were the guaianolides 10-epi-8-deoxycumambrin Apiaceae, Euphorbiaceae, Lauraceae, Magnoliaceae, Menis- B (compound 1), dehydroleucodin (compound 2) and ludartin permaceae, Rutaceae, Winteraceae and the Hepatidae. Bio- (compound 3). These compounds were competitive inhibitors with medical research has established that sesquiterpene lactones an apparent K{ = 4 [iM, K{ = 21 uM and K{ = 23 pJYl, respec- have potential utility in the treatment of solid tumors, stom- tively. Compounds 1 and 2 acted as type II ligands to the heme ach ulcers, antimigraines, cardiotonics [4] and more recently iron present in the active site of aromatase cytochrome P450 as inhibitors of the expression of cyclooxygenase [5] and the (P450arom). Besides, all of them failed to affect the cholesterol side-chain cleavage enzyme activity on human placental mito- transcription factor NF-KB [6]. No action on aromatase has chondrias. This is the first report on the aromatase inhibitory been so far described for the sesquiterpene lactones. In this activity of this group of natural compounds. work we report and characterise the inhibitory effect of this © 1997 Federation of European Biochemical Societies. group of compounds on human placental aromatase activity.

Key words: Aromatase inhibitor; Sesquiterpene lactone; 2. Materials and methods Type II ligand; Cholesterol side-chain cleavage; Human placental microsome; Argentine; Asteraceae 2.1. Materials The sesquiterpene lactones employed in the present study have been isolated and characterised as previously reported. Hence, they were 10-epi-8-deoxycumambrin B, ludartin [7], dehydroleucodin [8], peru- 1. Introduction vin, psilostachyin C, helenalin, psilostachyin, Sy II [9], achalensolide [10], Sy I [11] and eupahakonenin B [12], 3 14 The biosynthesis of estrogens is catalysed by the enzyme [l,2,6,7- H]Testosterone (80 Ci/mmol) and [4- C]Cholesterol (51 Ci/mmol) were purchased from New England Nuclear (Boston, complex aromatase. The complex is formed by a member of MA). Their purities were checked by thin layer chromatography. the superfamily of enzymes known collectively as cytochrome All other non-radioactive steroids, NADPH, DL-dithiothreitol, ethyl- P450, namely, aromatase cytochrome P450 (P450arom, the enediaminotetraacetic acid and other reagents were purchased from product of CYP19 gene). Associated with the P450arom is a Sigma Chemical Co. (St. Louis, MO). flavoprotein, the NADPH cytochrome P450 reductase. These two enzymes catalyse the aromatization of the A ring of an- 2.2. Biological preparations Human placental microsomes (particles sedimenting at 105 000 X g drogens to form the phenolic ring characteristic of estrogens. for 60 min) were obtained as reported [13] and human placental mi- This reaction involves three sequential oxidations to the C-19 tochondrias were obtained using the procedure described by Tuckey methyl group of the substrate, followed by elimination of C- [14]. The protein concentration was determined according to the Brad- 19 as formic acid and aromatization of the A ring. This re- ford procedure using BSA as standard [15]. action is the only one in vertebrates that introduces an aro- matic ring into a molecule [1]. Regulation of this enzyme plays 2.3. Assay of aromatase activity The enzyme activity was determined by measuring the conversion of an important role in many physiological processes and in [3H]testosterone to [3H]estradiol plus [3H]estrone as described earlier certain diseases, the most important of which is estrogen de- [13]. To test the effect of different compounds on the aromatase ac- pendent breast cancer. The use of aromatase inhibitors for tivity present in the microsomal fractions, varying amounts of them treatment of breast cancer is well demonstrated nowadays. were added to the microsomes from ethanolic stock solutions. Ethanol The several different kinds of inhibitors investigated include alone, which did not exceed a concentration of 1% was added to control incubations. Values of Km and Kmax were estimated graphi- steroidal and non-steroidal compounds which act in a com- cally from plots of l/V vs. US using linear regression analysis. The petitive or in a mechanism-based manner [2]. replots from the slope of each reciprocal plot versus the corresponding inhibitor concentrations were generated and the K\ values for these The plant kingdom is a very important source of medicinal compounds were determined.

»Corresponding author. Fax: (54) (51) 334174. 2.4. Assay of cholesterol side-chain cleavage enzyme activity E-mail: [email protected] The enzyme activity was determined by measuring the conversion of

1. lO-epi-8-deoxycumambrinB 2. Dehydroleucodin 3. Ludartin ICso-7nM IC50-15nM IC50= 55uM

4. Peruvin 5. Helenalin IC50=65nM ICso=70uM

7. Achalensolide 9 Syl lC5o>200>iM IC5o = llOuM

flfe*- -0*1

10. Eupahakonenin 6 11. Psilostachyin C IC5o>200nM IC50>200nM Fig. 1. Structures of sesquiterpene lactones and their IC50 values for aromatase activity in human placental microsomes. Values represent the average of duplicate determinations in two separate experiments. Substrate = 50 nM testosterone.

[14C]cholesterol to [14C]pregnenolone plus [14C]progesterone as de- ing standard graphical analysis. Values represent the mean of three scribed by Robinson et al. [16] with some modifications. The com- experiments. pounds were added to the mitochondrias from ethanolic stock solutions. Ethanol alone, which did not exceed a concentration of 1%, was 2.6. Molecular modelling added to the control incubations. After a preincubation time of 5 min The structures of the inhibitors and substrates were minimised using the reaction was started by adding 0.1 ml of a mixture of AMI calculations with MOPAC. [14C]cholesterol (80000 cpm/ml) sonicated in the presence of Tween 80 [17]. The final concentration of substrate was 1.5 uM and the final concentration of detergent was 20 ug/ml. The reaction was linear for 3. Results at least 4 h. Cholesterol side-chain cleavage enzyme activity was ex- 14 pressed as a percentage of radioactivity found in [ C]pregnenolone Eleven sesquiterpene lactones of different skeleton-type, 14 plus [ C]progesterone divided by the total radioactivity. The choles- shown in Fig. 1, were tested for aromatase inhibitory activity. terol side-chain cleavage enzyme inhibitory activity of the compounds was assessed in terms of 50% inhibitory concentration. None of the compounds were found to stimulate enzyme activity. Compounds 17 produced dose-dependent inhibition of aromatase with IC50 values ranging from 7 to 110 (0,M, while 2.5. Spectral studies compounds 8-11 were inactive (Fig. 1). Spectroscopic studies were carried out as described by Kellis and The most active compounds, 1-3 (10-epi-8-deoxycumam- Vickery [18]. Spectra were recorded with an UV 1601 PC/UV Shimad- zu spectrophotometer interfaced to an UV/PC personal spectroscopy brin B, dehydroleucodin and ludartin, respectively), belong software. The protein concentration was 6.0-10.0 mg/ml. Briefly, ar- to the same skeleton-type (guaianolide with trans-fusion of omatase cytochrome P450 was extracted from human placental micro- the y-lactone ring to C-6). For this reason we performed a somes with 0.5% sodium cholate in the presence of 1 uM of testoster- more detailed study of their inhibition mechanisms. Fig. 2A one. The detergent-solubilized enzyme was collected as a 55% saturated ammonium sulphate pellet and dialysed against 0.05 M shows the time course of the aromatization of 50 pmol of sodium phosphate buffer (pH 7.2). A baseline spectrum was recorded testosterone in 1.0 ml (100000 cpm/ml) by human placental with enzyme and substrate (0.25 uM testosterone) in the sample and microsomes in the presence and absence of compound 1. reference cells and then, different concentrations of the compounds Under these conditions, 10 u,M of the compound inhibited were added to the sample cell (solvent dimethyl sulfoxide alone was the aromatization by approximately 70% (Fig. 2A, inset). added to the reference). Repeated spectra were recorded to ensure sample equilibration and 3-5 spectra were routinely signal averaged. The reaction rate in the presence of the inhibitor remains The reported apparents binding constants (Kslipp) were calculated us- linear for 10 min. The lack of time-dependence of inhibition 398 J.G. Blanco et al.lFEBS Letters 409 (1997) 396^00

100.0nM 100

0 2 4 6 8 1012 Time (min)

10.0uM

5.0uM

inhibitor = 10.0(iM T~ f°" "l r 12 -.05 -.04 -.03 -.02 -.01 0.00 .01 1/[S] (nMol)-1 A B Fig. 2. (A) Time course of aromatization in microsomes in presence and absence of compound 1(10 uM). Samples contained 100 ug of protein and 50 pmol of testosterone in 1.0 ml. The inset shows the activity of the inhibited sample as percentage of that of the control at each time. (B) Kinetic analysis of inhibition of testosterone aromatization in human placental microsomes by compound 1. Inset, slopes of lines in double reciprocal plot vs. inhibitor concentration. Each point represents the mean of three experiments performed in duplicate. indicates that no significant conversion of the inhibitor to of 40 uM. Compound 3 showed a plain curve difference spec- more or less active forms, occurs during the incubation; sim- trum. When compound 1 was added to aromatase in absence ilar results were obtained with compounds 2 and 3. In addi- of the substrate (testosterone), no spectral changes were ob- tion, when human placental microsomes were pre-incubated in presence of 10 uM of compounds 1-3 and absence of substrate, variation of the pre-equilibration time failed to affect the fractional inhibition or time course (data not shown). Kinetic analyses were also performed to investigate the mechanism by which the sesquiterpene lactones inhibit aromatase enzyme activity. Fig. 2B shows a Lineweaver-Burk plot of the inhibition of human placental aromatase by compound 1. In- hibition was competitive with respect to the substrate testosterone and a replot of the slopes of the lines (shown in the inset) yielded a K[ value of 4 uM for compound 1. Similar results were obtained for compounds 2 (Äj = 21 uJVI) and 3 (Ki = 23 uM) (data not shown). In the absence of inhibitors, the Km of testosterone averaged 38 nM. The kinetic analyses suggested that compounds 1-3 may inhibit aromatization by competing with testosterone for the substrate binding site of the enzyme. This possibility was fur- ther investigated by monitoring the effects of these compounds on the absorption spectral properties of the enzyme. The difference spectra for compounds 1-3 together with that obtained for aminoglutethimide, a classical type II binder, are displayed in Fig. 3. When partially purified aromatase was -0.005 first equilibrated with testosterone, addition of compound 1 350.0 425.0 500.0 produced a type II difference spectrum with a Soret peak at 419 run. This type of spectrum reflected substrate displace- Wavelength (nm) ment and conversion of the high spin aromatase cytochrome Fig. 3. Difference spectra obtained upon interaction of compound 1 P450 to a low spin complex. Graphical analysis of the titra- [100 uM] (line 1), compound 2 [100 uM] (line 2), compound 3 [100 tion showed only one kind of binding site which exhibits a uM] (line 3) and aminoglutethimide [100 uM] (dotted line) with par- ^sapp of 29 u.M. A similar difference spectrum was obtained tially purified aromatase from human placenta. The samples contained 0.25 uM of testosterone and the protein concentration was for compound 2, showing a spectral binding constant CKsapp) 7.5 mg/ml. J.G. Blanco et al.lFEBS Letters 409 (1997) 396^00 399

served. This result is similar to that observed for 7,8-benzo- flavone, another natural non-steroidal inhibitor [18]. Some non-steroidal aromatase inhibitors exhibit dose-dependent effects on other cytochromes P450 mediated steroid hydroxylations. The cholesterol side-chain cleavage enzyme catalyses the three hydroxylations required for the conversion of cholesterol to pregnenolone. This enzyme is highly affected by aminoglutethimide [19]. We compared the capacity of aminoglutethimide and compounds 1 to 3 for inhibiting the ac- Compound 1 (10-epi-8-deoxycumarabrin B) tivity of this enzyme present in human placental mitochondrias. Aminoglutethimide inhibited this enzyme activity at 100 uM by 50% but no significant inhibitory activity was obtained for compounds 1-3 in the range of concentrations tested (1- 200 |J.M). Fig. 4 shows a dose-response curve for the activity of aminoglutethimide and compound 1.

4. Discussion

Several natural products inhibit the aromatase enzyme, but Compound 2 (Dehydroleucodin) Compound 3 (Ludartin) the most relevant inhibitors belong to the groups of flavo- noids [18,20], lignans [21] and tobacco alkaloids [22]. In this Fig. 5. Three-dimensional models of compound 1 (10 epi-8-deoxycu- report we show that sesquiterpene lactones exhibit competitive mambrin B), compound 2 (dehydroleucodin) and compound 3 (ludartin). inhibition of human placental aromatase activity in a range of concentrations similar to the other plant-derived inhibitors reported previously. C-4 (epoxide in compound 3). From the three-dimensional On the basis of the kinetic evidences, spectral data, molec- computer generated model of 1-3 shown in Fig. 5, it is pos- ular modelling of our compounds and previous work on sible to see that: (a) in compound 1 the OH group is above three-dimensional model of aromatase [23], we propose a the plane of the C-10 member ring, (b) in compound 2 the mechanism of action of compounds 1-3. These compounds oxygen appears as a carbonyl group attached to C-2 and are structurally related and differ on the substitution pattern aligned in the plane of the molecule and, (c) in compound 3 and oxidation state of carbons C-l, C-2, C-3, C-4, C-10, C-14 the oxygénation is below the plane of the molecule as an and C-l5. At first glance, we observed a relationship between epoxide group at C-3, C-4. These displacements of the oxy- the activity and the position of the oxygen atom at C-10 (OH génation in the guaianolide nucleus from top to bottom in the in compound 1), at C-2 (carbonyl in compound 2) and at C-3, plane of the molecule is accompanied by a decreasing inhibitory activity. The type II difference spectra obtained for compounds 1 and 2 and not for compound 3, indicate that compounds 1 and 2 have the ability to coordinate to the heme iron at the active site. Besides, according to their spectra, compound 1 coordinates better than compound 2. During the aromatization reaction, the substrate (androstenedione or testosterone) has been suggested to position in the active site in such a manner that the methyl group C-l9 is directed toward the heme iron, which directly participates in the three sequential oxidation of this methyl group. Besides, according to the aromatase model proposed by Graham-Lor- ence et al. [23], the substrate is fixed in the active site via interaction of the C-3 carbonyl group with K473 (lysine) and the 17-keto group of the androstenedione or the 17-OH of the testosterone to the H128 (histidine), stabilising the substrate and holding the C-l9 methyl near the heme iron. In our case, we superimposed the models of compound 1 and testosterone by overlapping by hand the carbonyl group of compound 1 and the carbonyl group at C-3 of testosterone. We thus found a combination of both structures in which the .1 1.0 10.0 100.0 hydroxyl group at C-10 of compound 1 is spatially close to the C-l9 methyl group of testosterone (Fig. 6). This proximity Inhibitor (|iM) in space would indicate that in analogy with the model pro- Fig. 4. Comparative effects of compound 1 and aminoglutethimide posed by Graham-Lorence et al. [23], the carbonyl group of on cholesterol side-chain cleavage enzyme activity (CSCC) present the lactone would interact with the K473 residue and the OH in human placental mitochondrias. The cholesterol concentration group at C-10 would coordinate to the heme iron during the was 1.5 u.M. Each point represents the mean of two experiments performed in duplicate. inhibition mechanism. In the case of compound 2, which 400 J. G. Blanco et al.lFEBS Letters 409 (1997) 396^00

Acknowledgements : This study was supported in part by Consejo Na- tional de Investigaciones Cientificas y Técnicas (CONICET), Consejo de Investigaciones Cientificas y Tecnolôgicas de la Provincia de Cor- doba (CONICOR), Fundaciôn Antorchas and Fundaciôn Alberto Roemmers. We wish to thank Dr. Oscar Giordano for his generous gift of dehydroleucodin.

References

[1] Simpson, E.R., Mahendroo, M.S., Means, G.D., Kilgore, M.W., Hinsheiwood, M.M., Graham-Lorence, S., Amarneh, B., Ito, Y., Fisher, C.R., Michael, M.D., Mendelson, C.R., Bulun, S.E., En- docr. Rev. 15 (1994) 342-355. [2] Brodie, A.M.H., J. Steroid Biochem. Mol. Biol. 49 (1994) 281- 287. Fig. 6. Three-dimensional models of compound 1 (dotted line) and [3] NAPRALERT (SM), (1996). Database College of Pharmacy of testosterone (full line). The models are superimposed by overlapping the University of Illinois at Chicago, 833 South Wood Street, by hand the carbonyl group at C-12 of compound 1 and the car- Chicago, IL 60612. bonyl group at C-3 of testosterone. S, steroid; L, sesquiterpene lac- [4] Robles, M., Aregullin, M., West, J., Rodriguez, E., Planta Med. tone. 61 (1995) 199-203. [5] Hwang, D., Fischer, N.H., Jang, B.C., Tak, H., Kim, J.K., Lee, shows a weaker type II difference spectra, the carbonyl oxygen W., Biochem. Biophys. Res. Commun. 226 (1996) 810-818. [6] Bork, P.M., Schmitz, M.L., Kuhnt, M., Escher, C, Heinrich, M., at C-2 seems to be responsible of coordinating to the heme FEBS Lett. 402 (1997) 85-90. iron. Considering that the coordination capacity of a ketone [7] Sosa, V.E., Oberti, J.C., Gil, R.R., Rüveda, E.A., Goedken, carbonyl is similar to an alcohol [24], the weaker type II V.L., Gutierrez, A.B., Herz, W., Phytochemistry 28 (1989) spectra and the weaker activity shown by compound 2 may 1925-1929. be due to the fact that carbonyl group at C-2 is farther from [8] Giordano, O.S., Guerreiro, E., Pestchanker, M.J., Guzman, J., Pastor, D., Guardia, T., J. Nat. Products 53 (1990) 803-805. the heme iron than the OH of compound 1. In the case of [9] Oberti, J.C., Silva, G.L., Sosa, V.E., Kulanthaivel, P., Herz, W., compound 3 the epoxide group is in the a face of the molecule Phytochemistry 25 (1986) 1355-1358. and close to the lactone carbonyl group, which may explain [10] Oberti, J.C., Sosa, V., Herz, W., Prasad, J., Goedken, V., J. Org. why this compound is less active than 1 and 2, and does not Chem. 48 (1983) 4038. present a type II spectra. On the other hand, the small size of [11] Gil, R.R., Pacciaroni del V., A., Oberti, J.C., Diaz, J.G., Herz, W., Phytochemistry 31 (1992) 593-596. these inhibitors compared with the steroidal nucleus seems to [12] Sosa, V.E., Gil, R.R., Oberti, J.C., Kulanthaivel, P., Herz, W., be a limitation for the possible interactions with the relatively J. Nat. Prod. 48 (1985) 340-341. large active site of the enzyme. [13] Genti Raimondi, S., Alvarez, CL, Patrito, L.C., Flury, A., J. Clin. Endocrinol. Metab. 76 (1993) 108-111. [14] Tuckey, R., J. Steroid Biochem. Mol. Biol. 42 (1992) 883-890. 5. Conclusions [15] Bradford, M.M., Anal. Biochem. 72 (1976) 248-254. [16] Robinson, J., Stevenson, P.M., Eur. J. Biochem. 24 (1971) 18-30. As a result of our investigation we can conclude that we are [17] Hochberg, R.B. vander Hoeven, TA., Welch, M., Lieberman, S., in presence of a new group of natural non-steroidal inhibitors Biochemistry 13 (1974) 603-609. [18] Kellis, J.T. Jr, Vickery, L.E., Science 225 (1984) 1032-1034. of the aromatase enzyme complex. Besides, the three most [19] Bhatnagar, A.S., Häusler, A., Schieweck, K., J. Enzyme Inhib- potent inhibitors here described (1, 2 and 3) have shown to ition 4 (1990) 179-186. be: (a) competitive with respect to the substrate testosterone; [20] Ibrahim, A.-R., Abul-Hajj, Y.J., J. Steroid Biochem. Mol. Biol. (b) more specific than aminoglutethimide, since they do not 37 (1990) 257-260. aifect the activity of the cholesterol side chain cleavage en- [21] Adlercreutz, H., Bannwart, C, Wähalä, K., Mäkela, T., Brunow, G., Hase, T., Arosemena, P.J., Kellis, J.T. Jr, Vickery, L.E., zyme, the first step of the steroidogenesis; (c) compounds 1 J. Steroid Biochem. Mol. Biol. 44 (1993) 147-153. and 2 act as type II ligands to the heme iron present in the [22] Barbieri, R.L., Gochberg, J., Ryan, K.J., J. Clin. Invest. 77 active site of aromatase cytochrome P450. (1986) 1727-1733. To our knowledge there are few guaianolides with a C-10 [23] Graham-Lorence, S., Amarneh, B., White, R.E., Peterson, JA., Simpson, E.R., Protein Sei. 4 (1995) 1065-1080. hydroxyl group ß-orientated and all were isolated from the [24] Dawson, J.H., Andersson, LA., Sono, M., J. Biol. Chem. 257 genus Stevia from north-western of Argentina [7,25]. This (1982) 3606-3617. uncommon stereochemistry opens interesting perspectives for [25] Sigstad, E.E., Catalan, C.A.N., Gutierrez, A.B., Diaz, J.G., the design of future modifications on the sesquiterpene nu- Goedken, V.L., Herz, W., Phytochemistry 30 (1991) 1933-1940. cleus in order to enhance its aromatase inhibitory activity.